Saxonica Developer Community

2024-05-09

• 
  Michael Kay Saxon

  20:53 · Closed issue

  I've applied a patch which I think makes the design more robust, so although we haven't got a test case to show that the problem is fixed, I'm going to mark it resolved (which means the patch will be in the next maintenance release). Bug #6323 (Resolved): Nullpointer Exception evaluating Java extension function call; suspected multithreading problem

  We are getting NullPointerException , see below. It doesnt give much information about which attribute or specific item it fails. Let us know what could be an issue or how to debug this issue. Can we get source code so we can debug code...
• 
  Michael Kay Further log files received by email today. No repro unfortunately, but something to be going on. My email response:
  OK, thanks for sending it.
  Excuse the fact that this note is a "thread of consciousness" - I'm exploring the proble...

  19:44 · Commented on issue
• 
  Michael Kay Saxon

  11:52 · Commented on issue

  I've been doing work on a fairly radical upheaval of line/column maintenance, and although much of this will probably only appear in the 13 branch, I'll report it here because some of if may find its way back into 12.x.
  Firstly, if line... Bug #6415: Saxon-HE 12.4J trace for unreferenced XSLT variables seems off

  The XSLT transformation below declares some global and local variables without referencing them. The trace behavior I'm seeing looks incorrect because the name of a trace element doesn't always match the line/column data.
  ### XSLT Sty...

2024-05-08

• 
  Michael Kay Saxon

  16:50 · Commented on issue

  I think this is almost right, but when we process a node in a collection, we should not only consider whether it has already been processed as part of the collection, but also whether it has already been processed for a different reason ... Bug #6406: A collection cannot contain the same document more than once

  The (infernal) constraint that `document-uri()`s must be unique shouldn't apply when it's the same document. See `src/test/java/s9apitest/TestCollections.java#testCollectionDupDoc`.
  The problem appears to arise from the fact that `Col...
• 
  Michael Kay Saxon

  12:24 · Closed issue

  Bug #6416 (Resolved): Documentation says saxon:compile-query and saxon:query are not available in SaxonCS but SaxonCS 12.4 seems to support them

  https://www.saxonica.com/html/documentation12/functions/saxon/compile-query.html says
  > Saxon availability
  > Requires Saxon-PE or Saxon-EE. Available for Java and C/C++ only (not .NET).
  >
  > Notes on the Saxon implementation
  > Ava...
• 
  Michael Kay I overlooked that since Saxon 11 we have supported `saxon:xquery()`, which implements the suggested design of returning a function item to evaluate the query.
  So: change of plan. We will change the documentation to mark the old functi...

  12:07 · Commented on issue
• 
  Michael Kay I think for the time being I'll go with documenting that these functions work under SaxonCS with known restrictions.

  11:47 · Edited issue
• 
  Michael Kay I've tried running our test cases for `saxon:compile-query()` and `saxon:query()` under SaxonCS and some of them fail, so I think the exclusion was deliberate.
  Looking at it more carefully, the tests that fail are those that explicitly ...

  10:55 · Commented on issue

2024-05-05

• 
  Michael Kay Saxon

  21:06 · Commented on issue

  What's happening here is a little bit strange. Saxon injects the trace instructions into the tree, then does optimization. The optimization removes the unused local variable declarations, but it doesn't remove the trace instructions, whi... Bug #6415: Saxon-HE 12.4J trace for unreferenced XSLT variables seems off

  The XSLT transformation below declares some global and local variables without referencing them. The trace behavior I'm seeing looks incorrect because the name of a trace element doesn't always match the line/column data.
  ### XSLT Sty...

2024-05-04

• 
  Michael Kay Saxon

  12:51 · Commented on issue

  Thanks. I'm afraid this is the kind of area where implementing the core feature - a different rounding mode - is trivial, but designing the APIs or configuration facilities to enable users to exploit the feature is a much bigger project.... Bug #6408: decimalDivide uses HALF_DOWN instead of HALF_UP rounding

  In class net.sf.saxon.expr.Calculator.java
  The function
  public static BigDecimalValue decimalDivide(NumericValue a, NumericValue b) throws XPathException
  is using HALF_DOWN instead of HALF_UP rounding, which would be the def...